The physics of tweets

They are studying the soundtrack of modern lives, made of posting, tweeting and liking.

Associate professor Joachim Mathiesen and his colleagues at the Niels Bohr Institute, University of Copenhagen, have discovered surprising analogies between how financial securities are traded, and how news is shared on Twitter, the popular microblogging site whose 200 million active users produce half a billion messages per day.

The findings have been recently reported in the scientific journal PNAS. The structure of prices and volatility are similar to the structure of surges in tweets and tweeting content.

Fifteen minutes of fame

Both on Twitter and on the stock market, most of the time there is just noise. Then something happens and attention is attracted by a company’s performance, a new publicity stunt or a politician’s press release. This causes activity bursts, with stocks changing hands more frequently than usual and people suddenly all tweeting on the same topic.

To track how interest on different firms is distributed over time, Professor Mathiesen measured the frequency by which brand names are mentioned on Twitter. He followed one hundred global brands like Apple, Coca-Cola and Samsung over a period of four months.

When a company launches a new product, people start to share opinions on that, and their tweets encourage others to do the same: “When a topic catches interest, people get activated by seeing other people involved and they start writing about the same thing because they feel related. It’s a self-amplifying positive feedback. That gives you a very characteristic peak structure” Prof. Mathiesen says to the University Post.

Fundamental human behaviour

Twitter Application Programming Interface (API), a free service that saves all messages containg a certain keyworld, together with their time stamp, in a text file.

“We did the same study tracking stocks, and we saw that trading volumes fluctuate in a similar way. This suggests that we are dealing with fundamental human behaviour, so that people get excited in the same way when something is happening, whether on Twitter or in the stock market,” he continues.

Politics and surges

The goal of the study is to investigate how people, in general, interact in massive communities. This could help, for example, to understand how political consensus is gained.

“If you are a politician with a lot of followers, you can create a lot of motion with your messages. When Obama says something, immediately a lot of discussion starts.”

Twitter had for example, a crucial role to play during the debate between Obama and Romney,” Mathiesen adds.

New fields

He is continuing his investigation into human behaviour by studying data collected by mobile phones, distributed among testers, as part of a collaboration between KU and DTU.

Analysis of tweets can also be applied to other spheres which involve humans. Researchers at the Niels Bohr Institute’s Centre for Models of Life use matemathical techniques to study complex systems, that range from the natural world, like viruses and ecological systems, to the social world.

Related to this field is the so-called ‘sentiment analysis’. Here scientists look at messages and statuses, classifying them as positive or negative for statistical analysis.

See a movie below made by Harvard University researchers showing how mood evolves during the day in the US:

Are people stressed by traffic? What’s their opinion on a movie, or a restaurant, or a politician?

All this, and more, can be gleaned from their tweets and status updates.

universitypost@adm.ku.dk

Like us on Facebook for features, guides and tips on upcoming events. Follow us on Twitter for links to other Copenhagen academia news stories. Sign up for the University Post weekly newsletter here.